Form Research - Project

Shemale

Hi all,

I have started working on my end of course college project and I am doing a racing form application, both codes and keeping it in GB and Ireland for the moment.

Once I have something working well I will send links to regular posters who PM looking, I will have a login with the idea being it will retain your preferences for research to save you some time.

Is there anything on the usual places that you like or don't like or anything you would like to see?

Ta.

PM me nudes

Options to filter form by ground and distance is a nice touch on ATR

Mr321

Like above the horses to the going. Also the class previous and headgear.

I'd imagine soon enough there should alot of information breakdown of previous headgear and what horse won wearing what.

Jockey trainer combo %

Good luck!

Shemale

Thanks lads, all good suggestions.

Gavin "shels"

Not sure if feasible - but at a snapshot the horses last win/winning weight listed at home. Unlike now you just get the previous form and finishing positions.

diomed

Like to see

(1) The real going, not the official going.
Almost every going description in flat racing had the word “good” in it.
My guess is clerks of the course are afraid to diverge from this.
Epsom for the English Derby is a race where they always use “good” in the description, but the times say different.
The real going can be calculated from race times, or from seconds per furlong.

(2) The soil type of the course: sandy, loamy, clay, chalky, peaty?
Soft going on one type of soil can be slower than soft on another soil type.
Unfortunately An Teagasc show “Urban” for soil types in towns, but you can figure out soil types by looking at their maps.
http://gis.teagasc.ie/soils/map.php

(3) An estimate of the best distance, going for each horse.
You can guess this by the number of times the horse raced at a distance (ignoring 2yo form), or the number of times a horse has run on the going.

(4) The fastest furlong time a horse has run.
This isn’t a timed furlong, but race times divided by race distance to give a seconds / furlong number.

(5) Good efforts from a bad draw and bad efforts from a good draw.
Horses unlikely to benefit from a good draw i.e. hold up horses.

diomed

(6) Wind direction and speed during the race for each race.
By wind direction I do not mean South, Southwest or stuff like that.

I want to know e.g. in the straight miles at Newmarket or The Curragh was the wind against or behind, how strong in miles per hour (of kilometers) and the angle.

A very good example of use of the wind was in the 2011 English 1000 Guineas which was run into a stiff breeze, against and from the right.
Frankie Dettori was drawn 16 of 18 which was on the right of the field, and in the wind that was coming from the right.
When the stalls opened he took a pull and brought the filly to the back of the field, and then took her across to the left of the pack, and moved forward a few lengths.
There he sheltered from the breeze, and inside the last furlong he has a fresh horse and won easily.
The first horses home were drawn 16,4,3,7,8,5,9 which shows that those drawn on the stand side out of the wind had a huge advantage.
(the horse drawn 1 "reluctant to race and immediately tailed off")

Francie Barrett

What is the goal of the project? Is the goal to produce a complete software product, or is it more towards applying some computer theory to solve a problem?

If it's the latter, I would say doing both flat and jump racing is a bad idea as the criteria you use for one won't be valid for the other. Also, I think it will be difficult to even get one code right, providing useful insights, let alone two. On top of that, a lecturer marking your project won't even care what code or geographies that you cover, so for a lot of effort you might get no credit.

I think what would interest a lecturer more is how you arrive to your insights. For example, I think picking a winner of a horse race is basically a classifier problem, and I would look to apply a classifier algorithm. In machine learning, one of the simpler probabilistic algorithms we can apply is naive Bayes. Once you gather your data, I would look to apply something like that to glean some analysis.

IanOBo

The 'has won of higher mark' indicator on irishracing.com is great

Shemale

Thanks Diomed, in terms of going do you mean the number? Plenty there for me to consider, not sure where I can get the information about winds and for soil types how dry loam compares to wet lime etc.

The deadline is 9 weeks away so I am focusing on readily available data to start with and persisted preferences for each user so the cards will render specific to the user.

Thanks Francie, I wont get the algorithms done by deadline but will definitely be trying my hand at predictive analysis. Two problems are sample size, it will take a while to get volume in the DB and each horse wont always be giving their true running which is a nightmare.

Thanks all for your input, that will keep me going, had considered a good bit of these features myself, just wanted to make sure I wasnt goung mad.

Gavin "shels"

As a matter interest - is it an IT course? Are you going to program from scratch?

diomed

Shemale wrote: »

Thanks Diomed, in terms of going do you mean the number?

I don't mean the number of horses going in the race.
I mean the state of the ground (the going).
Which of these is it: hard, firm; good to firm; good' good to soft; soft; sofo heavy; heavy?
And the problem is that almost always these going description are wrong.

Turftrax have going numbers based on readings from a "going stick"
http://www.turftrax.co.uk/going_maps.html

I realise many of my ideas are for information that is not readily available.
But there is no advantage to a gambler from readily available information / data.
You get an edge by using info that others do not use (or have).

diomed

A way to figure out the soil type goes something like this -
(the only really difficult soil type is clay)

To get a good bet in any race you are looking for extreme conditions.
These can be extreme distances, or terrain (big uphill slopes), or very heavy ground. Average conditions do little to separate horses.

How to figure out a course that has extremes of going - very fast ground and very slow ground (or very firm and very heavy).
Obviously the very fast ground is when it is very dry, and the very slow ground is when it is very wet.
The easy way to do this is to find out the big race(s) that are run at each course.
Go to the race results of that race(s) on Wikipedia and sort the race times from quickest to slowest.
A course with going extremes will have much bigger time ranges from fastest to slowest, often by as much as 10+ seconds over 12f.
You can do things like oppose a horse that won of soft on a chalky course when it runs on soft on a clay course.

This is a lot of work, and probably too much for your project, but it is something to consider for those serious about finding an edge.

Perhaps I should prepare for the flat season properly and do this for myself.

posturingpat

Trainer strikerate in that particular type of race. Claimer, 2yo maiden, 4yo hurdle etc. Races like claimers in particular get targetted and won by certain trainers.
Another interesting angle that i suspect would be extremely difficult to collaborate would be expected improvement for each trainer with horses from other yards first or second time out.

danganabu

diomed wrote: »

I don't mean the number of horses going in the race.
I mean the state of the ground (the going).
Which of these is it: hard, firm; good to firm; good' good to soft; soft; sofo heavy; heavy?

I'm pretty sure that shemale knew what you meant, and that he was asking did you mean the number on the going stick as opposed to the generic descriptions. Its a more accurate description but not sure if its available for all meetings.

danganabu

Shemale for ideas on possible reports I suggest you check out http://www.geegeez.co.uk they have some excellent reports there - I have been recording results from it for over 2 years and the most profitable ones are Horse for Courses and Handicap Debutants. I'm not a flat racing man but I have seen figures for the 2yo first time out report and they are also very impressive.

The most useful tool on those reports however are the filters and knowing how to use them, blindly following like most tools/systems etc will not give positive results.

Shemale

Gavin shels wrote: »

As a matter interest - is it an IT course? Are you going to program from scratch?

Sorry, just saw this now.

Yeah I am doing a H Dip in Software Development and this is final project, I will continue developing when the course is over.